CT3 April 2014 Q6 ii)

[Jamie]

Hi all,

I was wondering if someone could explain why we use a contingency table in part ii and not a standard proportion test. The two samples look to be independent so the only thing I can think of is that we are looking for a sample and not population proportion?

The conclusion from both tests is the same but want to know how we can spot when to use a contingency table in the exam - I understand if there is more than two samples we use them but for a simple comparison the two sample proportions I am not sure why we need the extra hassle.

Any help would be appreciated.

Cheers
Jamie

 

[John Lee]

In the case where there are only two outcomes, then a contingency table test is actually equivalent to the two sample proportions test. However, in this case we want to check that the proportions are unchanged for all three groups. So they would not be the same. Sure you could use proportions to show that, say, voting for candidate A vs not candidate A are the same - but it's not considering the votes between the other two candidates.